Authors
Evan Z. Macosko, Anindita Basu, ...,
Aviv Regev, Steven A. McCarroll
Correspondence
emacosko@genetics.med.harvard.edu
(E.Z.M.),
mccarroll@genetics.med.harvard.edu
(S.A.M.)
In Brief
Capturing single cells along with sets of
uniquely barcoded primer beads together
in tiny droplets enables large-scale,
highly parallel single-cell transcriptomics.
Applying this analysis to cells in mouse
retinal tissue revealed transcriptionally
distinct cell populations along with
molecular markers of each type.
Highlights
d
Accession Numbers
GSE63473
Resource
Highly Parallel Genome-wide Expression Profiling
of Individual Cells Using Nanoliter Droplets
Evan Z. Macosko,1,2,3,* Anindita Basu,4,5 Rahul Satija,4,6,7 James Nemesh,1,2,3 Karthik Shekhar,4 Melissa Goldman,1,2
Itay Tirosh,4 Allison R. Bialas,8 Nolan Kamitaki,1,2,3 Emily M. Martersteck,9 John J. Trombetta,4 David A. Weitz,5,10
Joshua R. Sanes,9 Alex K. Shalek,4,11,12 Aviv Regev,4,13,14 and Steven A. McCarroll1,2,3,*
1Department
SUMMARY
work, it will be important to learn the functional capacities and responses of each cell type.
A major determinant of each cells function is its transcriptional
program. Recent advances now enable mRNA-seq analysis of
individual cells (Tang et al., 2009). However, methods of preparing cells for profiling have been applicable in practice to just hundreds (Hashimshony et al., 2012; Picelli et al., 2013) or (with automation) a few thousand cells (Jaitin et al., 2014), typically after
first separating the cells by flow sorting (Shalek et al., 2013) or
microfluidics (Shalek et al., 2014) and then amplifying each cells
transcriptome separately. Fast, scalable approaches are needed
to characterize complex tissues with many cell types and states,
under diverse conditions and perturbations.
Here, we describe Drop-seq, a method to analyze mRNA
expression in thousands of individual cells by encapsulating
cells in tiny droplets for parallel analysis. Dropletsnanoliterscale aqueous compartments formed by precisely combining
aqueous and oil flows in a microfluidic device (Thorsen et al.,
2001; Umbanhowar et al., 2000)have been used as tiny reaction chambers for PCR (Hindson et al., 2011; Vogelstein and
Kinzler, 1999) and reverse transcription (Beer et al., 2008). We
sought here to use droplets to compartmentalize cells into nanoliter-sized reaction chambers for analysis of all of their RNAs. A
basic challenge of using droplets for transcriptomics is to retain
a molecular memory of the identity of the cell from which each
mRNA transcript was isolated. To accomplish this, we developed
a molecular barcoding strategy to remember the cell-of-origin of
each mRNA. We critically evaluated Drop-seq, then used it to
profile cell states along the cell cycle. We then applied it to a complex neural tissue, mouse retina, and from 44,808 cell profiles
identified 39 distinct populations, each corresponding to one or
a group of closely related cell types. Our results demonstrate
how large-scale single-cell analysis can help deepen our understanding of the biology of complex tissues and cell populations.
Complex tissue
Cell isolation
Cell suspension
STAMPs
Library
Single-cell transcriptomes
attached to microparticles
Synthesis
Round 1
UMI
Synthesis
Round 2
A
G
C
T
TTT(T27)
PCR
Cell
handle barcode
Synthesis
Round 12
A
G
C
T
4
CT
AG
8 rounds
of synthesis
16,777,216
48 different molecular
barcodes (UMIs) per bead
RESULTS
Drop-seq consists of the following steps (Figure 1A): (1) prepare
a single-cell suspension from a tissue; (2) co-encapsulate each
cell with a distinctly barcoded microparticle (bead) in a nanoliter-scale droplet; (3) lyse cells after they have been isolated in
droplets; (4) capture a cells mRNAs on its companion microparticle, forming STAMPs (single-cell transcriptomes attached to
microparticles); (5) reverse-transcribe, amplify, and sequence
thousands of STAMPs in one reaction; and (6) use the STAMP
barcodes to infer each transcripts cell of origin.
A Split-Pool Synthesis Approach to Generate Large
Numbers of Distinctly Barcoded Beads
To deliver large numbers of distinctly barcoded primer molecules into individual droplets, we use microparticles (beads).
We synthesized oligonucleotide primers directly on beads
(from 50 to 30 , yielding free 30 ends available for enzymatic priming). Each oligonucleotide is composed of four parts (Figure 1B):
(1) a constant sequence (identical on all primers and beads) for
use as a priming site for downstream PCR and sequencing; (2)
a cell barcode (identical across all the primers on the surface
of any one bead, but different from the cell barcodes on other
beads); (3) a Unique Molecular Identifier (UMI) (different on
each primer, to identify PCR duplicates) (Kivioja et al., 2012);
and (4) an oligo-dT sequence for capturing polyadenylated
mRNAs and priming reverse transcription.
Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc. 1205
field et al., 2002). We identified 544 human and 668 mouse genes
with expression patterns that varied along the cell cycle (at a
false discovery rate of 5%; Experimental Procedures) (Figure 4B),
including 200 orthologous gene pairs (p < 10!65 by hypergeometric test). Of these orthologous gene pairs, most (82.5%)
A
G1/S
G1/S
G2/M
G2/M
M/G1
M/G1
1
2
3
2
1
0
1
2
Avg. normalized
expression
Gene cluster
Phase-specific
score
3
4
5
5
6
7
8
8
50
150
250
350
450
550
50
C
Classic cell cycle genes
CCNB1
CCNB2
MCM2
MCM3
MCM4
MCM5
MCM6
MCM7
MCM10
AURKA
AURKB
100
150
200
250
300
ATF4
ARHGAP11A
ARPC2
CDCA4
E2F7
HISTH1E
MCMBP
NCAPG
NXT1
OTUB1
PARPBP
RPL26
SNHG3
SRP9
TCF19
WDHD1
ZFHX4
350
400
Embedding (tSNE) (Amir et al., 2013; van der Maaten and Hinton,
2008). We projected the remaining 36,145 cells in the data into
the tSNE analysis. We then combined a density clustering
approach with post hoc differential expression analysis to divide
44,808 cells among 39 transcriptionally distinct clusters (Supplemental Experimental Procedures) ranging from 50 to 29,400
cells in size (Figures 5B and 5C). Finally, we organized the 39
cell populations into larger categories (classes) by building a
dendrogram of similarity relationships among the 39 cell populations (Figure 5D, left).
The cell populations inferred from this analysis were readily
matched to the known retinal cell types, including all five
neuronal cell classes, based on the specific expression of known
markers for these cell types (Figure 5D, right, and Figure S6A).
Additional clusters corresponded to astrocytes (associated
with retinal ganglion cell axons exiting the retina), resident microglia, endothelial cells (from intra-retinal vasculature), pericytes,
and fibroblasts (Figure 5D). The relative abundances of the
major cell classes in our data agreed with earlier estimates
from microscopy (Jeon et al., 1998) (Table 1).
Replication and Cumulative Power of Drop-Seq Data
Replication across experimental sessions enables the construction of cumulatively powerful datasetsbut only if data are replicable and comparable. The retinal STAMPs were generated on 4
different days (weeks apart), utilizing different litters and multiple
runs in several sessions, for a total of seven replicates. One of the
Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc. 1207
Figure 5. Ab Initio Reconstruction of Retinal Cell Types from 44,808 Single-Cell Transcription Profiles Prepared by Drop-Seq
(A) Schematic representation of major cell classes in the retina. Photoreceptors (rods or cones) detect light and pass information to bipolar cells, which in turn
contact retinal ganglion cells that extend axons into other CNS tissues. Amacrine, bipolar and horizontal cells are retinal interneurons; Muller glia act as support
cells for surrounding neurons.
(B) Clustering of 44,808 Drop-seq single-cell expression profiles into 39 retinal cell populations. The plot shows a two-dimensional representation (tSNE) of global
gene expression relationships among 44,808 cells; clusters are colored by cell class, according to Figure 5A.
(C) Differentially expressed genes across 39 retinal cell populations. In this heat map, rows correspond to individual genes found to be selectively upregulated in
individual clusters (p < 0.01, Bonferroni corrected); columns are individual cells, ordered by cluster (139). Clusters with > 1,000 cells were downsampled to 1,000
cells to prevent them from dominating the plot.
(D) Gene expression similarity relationships among 39 inferred cell populations. Average expression across all detected genes was calculated for each of 39 cell
clusters, and the relative (Euclidean) distances between gene-expression patterns for the 39 clusters are represented by a dendrogram. The branches of the
dendrogram were annotated by examining the differential expression of known markers for retina cell classes and types. Twelve examples are shown at right,
using violin plots to represent the distribution of expression within the clusters. Violin plots for additional genes are in Figure S6A.
1208 Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc.
Cell Class
Percentage of Retina
(Jeon et al., 1998) (%)
Percentage of
Cell Population
in Drop-Seq (%)
Rod photoreceptors
79.9
65.6
Cone photoreceptors
2.1
4.2
Muller glia
2.8
3.6
0.5
1.0
Horizontal cells
0.5
0.6
Amacrine cells
7.0
9.9
Bipolar cells
7.3
14.0
Microglia
Astrocytes
0.2
0.6
0.1
The sizes of the 39 annotated cell clusters produced from Drop-seq were
used to estimate their fractions of the total cell population. These data
were compared with those obtained by microscopy techniques (Jeon
et al., 1998).
(E) Representation of experimental replicates in each cell population. tSNE plot from Figure 2B, with each cell now colored by experimental replicate (for visual
clarity, the central rod cluster was downsampled to 10,000 cells). Each of the seven replicates contributes to all 39 cell populations. Cluster 36 (arrow), in which
these replicates are unevenly represented, expressed markers of fibroblasts, which are not native to the retina and are presumably a dissection artifact (see also
Figure S6B).
(F) Trajectory of amacrine clustering as a function of number of cells analyzed. Three different downsampled datasets were generated: (1) 500, (2) 2,000, or (3)
9,731 cells (Supplemental Experimental Procedures). Cells identified as amacrines (clusters 323) in the full analysis are here colored by their cluster identities in
that analysis. Analyses of smaller numbers of cells incompletely distinguished these subpopulations from one another.
Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc. 1209
Figure 6. Finer-Scale Expression Distinctions among Amacrine Cells, Cones, and Retinal Ganglion Cells
(A) Pan-amacrine markers. The expression levels of the six genes identified (Nrxn2, Atp1b1, Pax6, Slc32a1, Slc6a1, Elavl3) are represented as dot plots across all
39 clusters; larger dots indicate broader expression within the cluster; deeper red denotes a higher expression level.
1210 Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc.
the MitoP line, validating this as a marker of nGnG cells (Figure 6F).
PPP1R17 was one of several markers that distinguished Cluster
20 from its closest neighbor, Cluster 21 (Figure 6G; 12 genes >
2.8-fold enrichment, p < 10!9). The differences between Clusters
20 and 21 suggest a hitherto unsuspected level of heterogeneity
among nGnG amacrines.
Supervised Analysis Reveals Additional Diversity
Our unsupervised analysis grouped cells into 39 transcriptionally distinct populations, but morphological and functional
criteria suggest that there are "100 retinal cell types. We asked
whether supervised analysis could reveal multiple types within
individual clusters. For example, retinal ganglion cells (RGCs),
which consist of about 30 types (Sanes and Masland, 2015),
formed a single cluster in our analysis, perhaps because it is
a rare cell population (1%, Table 1). Five RGC types, called
intrinsically photosensitive RGCs (ipRGCs), express Opn4, the
gene encoding the photopigment melanopsin. Opn4+ RGCs
(26/432) expressed nine genes at levels 2-fold higher than
Opn4- RGCs (p < 109, Figure 6H), including Tbr2/Eomes,
known to be a selective marker for this population (Sweeney
et al., 2014). This result reveals additional heterogeneity that
may also emerge ab initio as analyses expand to include
more cells.
DISCUSSION
Ascertaining transcriptional variation across individual cells is a
valuable way of learning about complex tissues and functional
responses, but single-cell analysis has been limited by the time
and cost of preparing libraries from many individual cells. A scientist employing Drop-seq can prepare 10,000 single-cell libraries for sequencing in 12 hr, for about 6.5 cents per cell (Table
S5), representing a >100-fold improvement in both time and cost
relative to existing methods. A Drop-seq setup can be constructed quickly and inexpensively in a standard biology lab using readily available equipment (Figure S2B and Supplemental
Experimental Procedures). We hope that ease, speed, and low
cost facilitate exuberant experimentation, careful replication,
and many cycles of experiments, analyses, ideas, and more
experiments.
In validating Drop-seq, we developed stringent species-mixing experiments to measure single-cell purity and cell doublet
rates in our libraries. In another article in this issue, Klein et al.
(Klein et al., 2015) describe a droplet-based approach to single-cell RNA-seq and also use species-mixing experiments to
evaluate it. Our results indicate that all methods of isolating
single cells from a cell suspension, including Drop-seq, fluorescence activated cell sorting (FACS) and microfluidics, are vulnerable to impurities, and highlight the value of performing species
mixing experiments to assess single-cell approaches. In our
retina analysis, even relatively impure libraries generated in
ultra-high-throughput modes (100 cells per ml, allowing the
processing of 10,000 cells per hour at "10% doublet and impurity rates) appeared to yield a robust and biologically validated
cell classification, but other tissues or applications may require
using Drop-seq in purer modes.
Unsupervised computational analysis of Drop-seq data
identified 39 transcriptionally distinct retinal cell populations,
many representing specific subtypes of the major retinal cell
classes (Figures 5 and 6). It is a particular strength of the
retina that establishing correspondence between cluster and
type was in many cases straightforward; an important direction will be to identify cell types and states in other parts of
the brainas well as in other tissuesabout which less is
currently known.
We see many applications of Drop-seq, beyond the identification of cell types and cell states. Genome-scale genetic studies
are identifying many genes whose variation contributes to disease
risk, but biology has lacked similarly high-throughput ways of
connecting these genes to specific cell populations and unique
functional responses. Drop-seq could be used to provide initial insights into how these genes function in the diverse cell types
composing each tissue. In addition, coupling Drop-seq to perturbationssuch as small molecules, mutations, pathogens, or other
stimulicould generate an information-rich, multi-dimensional
readout of the influence of perturbations on many kinds of cells.
The functional implications of a genes expression are a product not just of that genes intrinsic properties, but also of the
entire cell-level context in which the gene is expressed. We
hope Drop-seq enables the abundant and routine discovery of
such relationships in many areas of biology.
(B) Identification of known amacrine types among clusters. The 21 amacrine clusters consisted of 12 GABAergic, five glycinergic, one glutamatergic, and three
non-GABAergic non-glycinergic clusters. Starburst amacrines were identified in cluster 3 by their expression of Chat; excitatory amacrines by expression of
Slc17a8; A-II amacrines by their expression of Gjd2; and SEG amacrine neurons by their expression of Ebf3.
(C) Nomination of novel candidate markers of amacrine subpopulations. Each cluster was screened for genes differentially expressed in that cluster relative to all
other amacrine clusters (p < 0.01, Bonferroni corrected) (McDavid et al., 2013), and filtered for those with highest relative enrichment. Expression of a single
candidate marker for each cluster is shown across all amacrines.
(D) Validation of MAF as a marker for a GABAergic amacrine population. Staining of a fixed adult retina from wild-type mice for MAF (i, ii, v, and green staining in iv
and vii), GAD1 (iii and iv, red staining), and SLC6A9 (vi and vii, red staining), demonstrating co-localization of MAF with GAD1, but not SLC6A9.
(E) Differential expression of cluster 7 (Maf+) with nearest neighboring amacrine cluster (#6). Average gene expression was compared between cells in clusters 6
and 7; 16 genes (red dots) were identified with >2.8-fold enrichment in cluster 7 (p < 10!9).
(F) Validation of PPP1R17 as a marker for an amacrine subpopulation. Staining of a fixed adult retina from Mito-P mice, which express CFP in both nGnG
amacrines and type 1 bipolars (Kay et al., 2011). Overlapping labeling by PPP1R17 antibody (green) and Mito-P CFP (red) supports Drop-seq identification of
Ppp1r17 expression in the nGnG amacrine neurons. 85% of CFP+ cells were PPP1R17+ and 50% of the PPP1R17+ cells were CFP!, suggesting a second
amacrine type expressing this marker. Blue staining is for VSX2, a marker of bipolar neurons.
(G) Differential expression of cluster 20 (Ppp1r17+) with nearest neighboring amacrine cluster (#21). Average gene expression was compared between cells in
clusters 20 and 21; 12 genes (red dots) were identified with >2.8-fold enrichment in cluster 20 (p < 10!9).
(H) Differential expression of melanopsin-positive and negative RGCs. Average expression was compared between Opn4-positive and -negative RGCs in cluster
2. Seven genes were identified as enriched in Opn4-positive cells (red dots, > 2-fold, p < 10!9).
Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc. 1211
EXPERIMENTAL PROCEDURES
Device Design and Fabrication
Microfluidic devices were designed using AutoCAD software (Autodesk), and
the components tested using COMSOL Multiphysics (COMSOL). Full details
are described in Supplemental Experimental Procedures.
Barcoded Microparticle Synthesis
Bead functionalization and reverse-direction phosphoramidite synthesis were
performed by Chemgenes Corp (Wilmington, MA). Split-and-pool cycles
were accomplished by removing the dry resin from each column, hand mixing,
and weighing out four equal portions before returning the resin for an additional
cycle of synthesis. Full details are described in Supplemental Experimental
Procedures.
Drop-Seq Procedure
Monodisperse droplets "1 nl in size were generated using the microfluidic device described in Supplemental Experimental Procedures, in which barcoded
microparticles, suspended in lysis buffer, were flowed at a rate equal to that of
a single-cell suspension, so that resulting droplets were composed of an equal
amount of each component. As soon as droplet generation was complete,
droplets were broken with perfluorooctanol in 30 ml of 63 SSC. The addition
of a large aqueous volume to the droplets reduces hybridization events after
droplet breakage, because DNA base pairing follows second-order kinetics
(Britten and Kohne, 1968; Wetmur and Davidson, 1968). The beads were
then washed and resuspended in a reverse transcriptase mix, followed by a
treatment with exonuclease I to remove unextended primers. The beads
were then washed, counted, aliquoted into PCR tubes, and PCR amplified.
The PCR reactions were purified and pooled, and the amplified cDNA quantified on a BioAnalyzer High Sensitivity Chip (Agilent). The cDNA was fragmented and amplified for sequencing with the Nextera XT DNA sample prep
kit (Illumina) using custom primers that enabled the specific amplification of
only the 30 ends (Table S6). The libraries were purified, quantified, and then
sequenced on the Illumina NextSeq 500. All details regarding reaction
conditions, primers used, and sequencing specifications can be found in the
Supplemental Experimental Procedures.
Cell-Cycle Analysis of HEK and 3T3 Cells
Gene sets reflecting five phases of the HeLa cell cycle (G1/S, S, G2/M, M and
M/G1) were taken from Whitfield et al. (Whitfield et al., 2002) with some modification (Supplemental Experimental Procedures and Table S2). A phase-specific score was generated for each cell, across all five phases, using averaged
normalized expression levels (log2(TPM+1)) of the genes in each set. Cells
were then ordered along the cell cycle by comparing the patterns of these
five phase scores per cell. To identify cell-cycle-regulated genes, we used a
sliding window approach, and identified windows of maximal and minimal
average expression, both for ordered cells, and for shuffled cells, to evaluate
the false-discovery rate. Full details may be found in Supplemental Experimental Procedures.
Principal Components and Clustering Analysis of Retina Data
The clustering algorithm for the retinal cell data was implemented and performed using Seurat, a recently developed R package for single-cell analysis
(Satija et al., 2015). PCA was first performed on a 13,155-cell training set
of the 49,300-cell dataset, using single-cell libraries in which transcripts from
>900 genes were detected. We found this approach was more effective in
discovering structures corresponding to rare cell types than performing PCA
on the full dataset, which was dominated by numerous, tiny rod photoreceptors (Supplemental Experimental Procedures). Thirty-two statistically significant PCs were identified using a permutation test and independently
confirmed using a modified resampling procedure (Chung and Storey, 2015).
We projected individual cells within the training set based on their PC scores
onto a single two-dimensional map using t-Distributed Stochastic Neighbor
Embedding (t-SNE) (van der Maaten and Hinton, 2008). The remaining
36,145 single-cell libraries (<900 genes detected) were next projected on
this t-SNE map, based on their representation within the PC-subspace of
the training set (Berman et al., 2014; Shekhar et al., 2014). This approach mit-
1212 Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc.
igates the impact of noisy variation in the lower complexity libraries due to
gene dropouts. It was also reliable in the sense that when we withheld from
the t-SNE all cells from a given cluster and then tried to project them, these
withheld cells were not spuriously assigned to another cluster by the projection
(Table S7). Point clouds on the t-SNE map represent candidate cell types; density clustering (Ester et al., 1996) identified these regions. Differential expression testing (McDavid et al., 2013) was then used to confirm that clusters
were distinct from each other. Hierarchical clustering based on Euclidean distance and complete linkage was used to build a tree relating the clusters. We
noted expression of several rod-specific genes, such as Rho and Nrl, in every
cell cluster, an observation that has been made in another retinal cell gene
expression study (Siegert et al., 2012) and likely arises from solubilization
of these high-abundance transcripts during cell suspension preparation.
Additional information regarding retinal cell data analysis can be found in the
Supplemental Experimental Procedures.
ACCESSION NUMBERS
The accession number for the raw and analyzed data reported in this paper is
GEO: GSE63473.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, seven tables, one movie, and one data file and can be found with
this article online at http://dx.doi.org/10.1016/j.cell.2015.05.002.
AUTHOR CONTRIBUTIONS
E.Z.M. developed the barcoding and molecular biology analysis, advised by
S.A.M. A.B. designed and fabricated the microfluidic devices, advised by
D.A.W. and A.R. E.Z.M. and M.G. developed Drop-seq experimental protocols
and performed the Drop-seq experiments in S.A.M.s lab. J.N. developed the
methods and software for obtaining digital gene expression measurements for
each cell, advised by E.Z.M. and S.A.M. J.N., E.Z.M. and S.A.M. performed the
analyses of species-mixing experiments. I.T. performed the cell-cycle analysis. A.R.B. prepared the retinal cell suspensions. R.S., K.S., and A.R. developed and performed the retinal cell type clustering analyses with contribution
from N.K. E.Z.M., R.S., K.S., and J.R.S. interpreted the retina expression data.
E.M.M. and J.R.S. performed the immunohistochemistry experiments. J.J.T.
and A.K.S. performed the Fluidigm C1 experiments. E.Z.M., S.A.M., A.R.,
A.B., and A.K.S. conceived the study and key ways that Drop-seq works
together as an integrated system. E.Z.M. and S.A.M. wrote the manuscript
with contributions from all authors.
ACKNOWLEDGMENTS
This work was supported by the Stanley Center for Psychiatric Research (to
S.M.), the MGH Psychiatry Residency Research Program and Stanley-MGH
Fellowship in Psychiatric Neuroscience (to E.Z.M.), a Stewart Trust Fellows
Award (to S.M.), a grant from the Simons Foundation to the Simons Center
for the Social Brain at MIT (to A.R., S.M., and D.W.), an NHGRI CEGS P50
HG006193 (to A.R.), the Klarman Cell Observatory (to A.R. and A.B.), NIMH
grant U01MH105960 (to S.M., A.R. and J.R.S.), NIMH grant R25MH094612
(to E.M.), NIH F32 HD075541 (to R.S.). AR is an investigator of the Howard
Hughes Medical Institute. Microfluidic device fabrication was performed at
the Harvard Center for Nanoscale Systems (CNS), a member of the National
Nanotechnology Infrastructure Network (National Science Foundation award
no. ECS-0335765), with support from the National Science Foundation
(DMR-1310266) and the Harvard Materials Research Science and Engineering
Center (DMR-1420570). We thank Christina Usher and Leslie Gaffney for contributions to the manuscript figures and Chris Patil for helpful comments on the
manuscript. We thank Connie Cepko for helpful conversations about the retina
data, Beth Stevens for advice on retinal dissociations, and Assaf Rotem and
Huidan Zhang for advice on microfluidics design and fabrication. A.R. is a
member of the Scientific Advisory Board for Thermo Fisher Scientific and
Syros Pharmaceuticals and a consultant for Driver Genomics.
Kay, J.N., Voinescu, P.E., Chu, M.W., and Sanes, J.R. (2011). Neurod6 expression defines new retinal amacrine cell subtypes and regulates their fate. Nat.
Neurosci. 14, 965972.
Kivioja, T., Vaharautio, A., Karlsson, K., Bonke, M., Enge, M., Linnarsson, S.,
and Taipale, J. (2012). Counting absolute numbers of molecules using unique
molecular identifiers. Nat. Methods 9, 7274.
REFERENCES
Amir, A.D., Davis, K.L., Tadmor, M.D., Simonds, E.F., Levine, J.H., Bendall,
S.C., Shenfeld, D.K., Krishnaswamy, S., Nolan, G.P., and Peer, D. (2013).
viSNE enables visualization of high dimensional single-cell data and reveals
phenotypic heterogeneity of leukemia. Nat. Biotechnol. 31, 545552.
Beer, N.R., Wheeler, E.K., Lee-Houghton, L., Watkins, N., Nasarabadi, S.,
Hebert, N., Leung, P., Arnold, D.W., Bailey, C.G., and Colston, B.W. (2008).
On-chip single-copy real-time reverse-transcription PCR in isolated picoliter
droplets. Anal. Chem. 80, 18541858.
Berman, G.J., Choi, D.M., Bialek, W., and Shaevitz, J.W. (2014). Mapping the
stereotyped behaviour of freely moving fruit flies. J. R. Soc. Interface 11,
20140672.
Brennecke, P., Anders, S., Kim, J.K., Ko1odziejczyk, A.A., Zhang, X., Proserpio, V., Baying, B., Benes, V., Teichmann, S.A., Marioni, J.C., and Heisler,
M.G. (2013). Accounting for technical noise in single-cell RNA-seq experiments. Nat. Methods 10, 10931095.
Britten, R.J., and Kohne, D.E. (1968). Repeated sequences in DNA. Hundreds
of thousands of copies of DNA sequences have been incorporated into the
genomes of higher organisms. Science 161, 529540.
Klein, A.M., Mazutis, L., Akartuna, I., Tallapragada, N., Veres, A., Li, V., Peshkin, L., Weitz, D.A., and Kirschner, M.W. (2015). Droplet barcoding for single
cell transcriptomics and its application to embryonic stem cells. Cell 161,
this issue, 11871201.
Luo, L., Callaway, E.M., and Svoboda, K. (2008). Genetic dissection of neural
circuits. Neuron 57, 634660.
Masland, R.H. (2012). The neuronal organization of the retina. Neuron 76,
266280.
McDavid, A., Finak, G., Chattopadyay, P.K., Dominguez, M., Lamoreaux, L.,
Ma, S.S., Roederer, M., and Gottardo, R. (2013). Data exploration, quality control and testing in single-cell qPCR-based gene expression experiments. Bioinformatics 29, 461467.
Petilla Interneuron Nomenclature Group, Ascoli, G.A., Alonso-Nanclares, L.,
Anderson, S.A., Barrionuevo, G., Benavides-Piccione, R., Burkhalter, A., Buzsaki, G., Cauli, B., Defelipe, J., Fairen, A., et al. (2008). Petilla terminology:
nomenclature of features of GABAergic interneurons of the cerebral cortex.
Nat. Rev. Neurosci. 9, 557568.
Picelli, S., Bjorklund, A.K., Faridani, O.R., Sagasser, S., Winberg, G., and
Sandberg, R. (2013). Smart-seq2 for sensitive full-length transcriptome
profiling in single cells. Nat. Methods 10, 10961098.
Sanes, J.R., and Masland, R.H. (2015). The Types of Retinal Ganglion Cells:
Current Status and Implications for Neuronal Classification. Annu. Rev. Neurosci. Published online April 9, 2015.
Descamps, F.J., Martens, E., Proost, P., Starckx, S., Van den Steen, P.E., Van
Damme, J., and Opdenakker, G. (2005). Gelatinase B/matrix metalloproteinase-9 provokes cataract by cleaving lens betaB1 crystallin. FASEB J. 19,
2935.
Sanes, J.R., and Zipursky, S.L. (2010). Design principles of insect and vertebrate visual systems. Neuron 66, 1536.
Ester, M., Kriegel, H.P., Sander, J., and Xu, X. (1996). A density-based algorithm for discovering clusters in large spatial databases with noise (Menlo
Park, Calif: AAAI Press).
Famiglietti, E.V., and Sundquist, S.J. (2010). Development of excitatory and
inhibitory neurotransmitters in transitory cholinergic neurons, starburst amacrine cells, and GABAergic amacrine cells of rabbit retina, with implications
for previsual and visual development of retinal ganglion cells. Vis. Neurosci.
27, 1942.
Feigenspan, A., Teubner, B., Willecke, K., and Weiler, R. (2001). Expression
of neuronal connexin36 in AII amacrine cells of the mammalian retina.
J. Neurosci. 21, 230239.
Hashimshony, T., Wagner, F., Sher, N., and Yanai, I. (2012). CEL-Seq: singlecell RNA-Seq by multiplexed linear amplification. Cell Rep. 2, 666673.
Haverkamp, S., and Wassle, H. (2004). Characterization of an amacrine cell
type of the mammalian retina immunoreactive for vesicular glutamate transporter 3. J. Comp. Neurol. 468, 251263.
Hindson, B.J., Ness, K.D., Masquelier, D.A., Belgrader, P., Heredia, N.J.,
Makarewicz, A.J., Bright, I.J., Lucero, M.Y., Hiddessen, A.L., Legler, T.C.,
et al. (2011). High-throughput droplet digital PCR system for absolute quantitation of DNA copy number. Anal. Chem. 83, 86048610.
Satija, R., Farrell, J.A., Gennert, D., Schier, A.F., and Regev, A. (2015). Spatial
reconstruction of single-cell gene expression data. Nat. Biotechnol. Published
online 13 April, 2015. http://dx.doi.org/10.1038/nbt.3192.
Shalek, A.K., Satija, R., Adiconis, X., Gertner, R.S., Gaublomme, J.T.,
Raychowdhury, R., Schwartz, S., Yosef, N., Malboeuf, C., Lu, D., et al.
(2013). Single-cell transcriptomics reveals bimodality in expression and
splicing in immune cells. Nature 498, 236240.
Shalek, A.K., Satija, R., Shuga, J., Trombetta, J.J., Gennert, D., Lu, D., Chen,
P., Gertner, R.S., Gaublomme, J.T., Yosef, N., et al. (2014). Single-cell
RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510,
363369.
Shekhar, K., Brodin, P., Davis, M.M., and Chakraborty, A.K. (2014). Automatic
Classification of Cellular Expression by Nonlinear Stochastic Embedding
(ACCENSE). Proc. Natl. Acad. Sci. USA 111, 202207.
Siegert, S., Cabuy, E., Scherf, B.G., Kohler, H., Panda, S., Le, Y.Z., Fehling,
H.J., Gaidatzis, D., Stadler, M.B., and Roska, B. (2012). Transcriptional code
and disease map for adult retinal cell types. Nat. Neurosci. 15, 487495,
S1S2.
Sweeney, N.T., Tierney, H., and Feldheim, D.A. (2014). Tbr2 is required to
generate a neural circuit mediating the pupillary light reflex. J. Neurosci. 34,
54475453.
Islam, S., Zeisel, A., Joost, S., La Manno, G., Zajac, P., Kasper, M., Lonnerberg, P., and Linnarsson, S. (2014). Quantitative single-cell RNA-seq with
unique molecular identifiers. Nat. Methods 11, 163166.
Tang, F., Barbacioru, C., Wang, Y., Nordman, E., Lee, C., Xu, N., Wang, X.,
Bodeau, J., Tuch, B.B., Siddiqui, A., et al. (2009). mRNA-Seq whole-transcriptome analysis of a single cell. Nat. Methods 6, 377382.
Jaitin, D.A., Kenigsberg, E., Keren-Shaul, H., Elefant, N., Paul, F., Zaretsky, I.,
Mildner, A., Cohen, N., Jung, S., Tanay, A., and Amit, I. (2014). Massively parallel single-cell RNA-seq for marker-free decomposition of tissues into cell
types. Science 343, 776779.
Thorsen, T., Roberts, R.W., Arnold, F.H., and Quake, S.R. (2001). Dynamic
pattern formation in a vesicle-generating microfluidic device. Phys. Rev.
Lett. 86, 41634166.
Jeon, C.J., Strettoi, E., and Masland, R.H. (1998). The major cell populations of
the mouse retina. J. Neurosci. 18, 89368946.
Umbanhowar, P.B., Prasad, V., and Weitz, D.A. (2000). Monodisperse Emulsion Generation via Drop Break Off in a Coflowing Stream. Langmuir 16,
347351.
Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc. 1213
Utada, A.S., Fernandez-Nieves, A., Stone, H.A., and Weitz, D.A. (2007).
Dripping to jetting transitions in coflowing liquid streams. Phys. Rev. Lett.
99, 094502.
van der Maaten, L., and Hinton, G. (2008). Visualizing Data using t-SNE.
J. Mach. Learn. Res. 9, 25792605.
Vogelstein, B., and Kinzler, K.W. (1999). Digital PCR. Proc. Natl. Acad. Sci.
USA 96, 92369241.
Wetmur, J.G., and Davidson, N. (1968). Kinetics of renaturation of DNA. J. Mol.
Biol. 31, 349370.
1214 Cell 161, 12021214, May 21, 2015 2015 Elsevier Inc.
Whitfield, M.L., Sherlock, G., Saldanha, A.J., Murray, J.I., Ball, C.A., Alexander,
K.E., Matese, J.C., Perou, C.M., Hurt, M.M., Brown, P.O., and Botstein, D.
(2002). Identification of genes periodically expressed in the human cell cycle
and their expression in tumors. Mol. Biol. Cell 13, 19772000.
Yang, Y., and Cvekl, A. (2005). Tissue-specific regulation of the mouse alphaAcrystallin gene in lens via recruitment of Pax6 and c-Maf to its promoter. J. Mol.
Biol. 351, 453469.
Zhu, Y.Y., Machleder, E.M., Chenchik, A., Li, R., and Siebert, P.D. (2001).
Reverse transcriptase template switching: a SMART approach for full-length
cDNA library construction. Biotechniques 30, 892897.